Mutually Permutable Products of two Nilpotent Groups.
A. BALLESTER-BOLINCHES(*) - J. C. BEIDLEMAN(**) - J. COSSEY(***) H. HEINEKEN(***) - M. C. PEDRAZA-AGUILERA(§)
Dedicated to Professor G. Zappa on the occasion of his ninetieth birthday
1. Introduction.
Throughout the paper only finite groups are considered. Two sub- groupsAandBof a groupGare calledmutually permutableif and only if AY YAandXBBX is true for allY BandXA; they are called totally permutable ifXY YX for allYBandXA. These two con- cepts are more general than those of normal products and central pro- ducts, for instance the dihedral group of order 6 is a product of two nil- potent totally permutable subgroups. In both cases, under some restric- tions on the structure of the factors, the product is a supersoluble group (see Carocca [7] or [6, Corollary]), a class of groups considered much earlier by Zappa (see [14] and [15]). Asaad and Shaalan [1] study sufficient conditions for totally and mutually permutable products of two super- soluble subgroups to be supersoluble. More precisely they prove:
Let G be the mutually permutable product of the supersoluble sub- groups A and B. If either the product is totally permutable or A or B is nilpotent,then G is supersoluble.
(*) Indirizzo dell'A.: Departament d'AÁlgebra, Universitat de ValeÁncia, Dr.
Moliner 50, 46100 Burjassot, ValeÁncia (Spain); e-mail: Adolfo.Ballester@uv.es (**) Indirizzo dell'A.: Department of Mathematics, University of Kentucky, Lexington, Kentucky 40506-0027 U.S.A.; e-mail: clark@ms.uky.edu
(***) Indirizzo dell'A.: Department of Mathematics, Mathematical Sciences Institute, Australian National University, Canberra ACT 0200, Australia; e-mail:
John.Cossey@maths.anu.edu.au
(***) Indirizzodell'A.:MathematischesInstitut,UniversitaÈtWuÈrzburg,AmHub- land, 97074 WuÈrzburg (Germany); e-mail: heineken@mathematik.uni-wuerzburg.de (§) Indirizzo dell'A.: Departamento de MatemaÂtica Aplicada, E.T.S.I.A, Uni- versidad PoliteÂcnica de Valencia, Camino de Vera, 46071, Valencia (Spain); e-mail:
mpedraza@mat.upv.es
The above results have been studied within the framework of formation theory. This was the beginning of an intensive study of this kind of pro- ducts, even in the infinite case (see [3,4,5,8]). Moreover Ballester-Bo- linches, Cossey and Esteban-Romero have shown that totally permutable products of nilpotent groups are extensions of an abelian by a nilpotent group (see [2, Theorem 1]). We are interested here to find more in- formation about the structure of the product of two nilpotent mutually permutable subgroups. In particular, if G is the mutually permutable product of two nilpotent subgroups of odd order, then we obtain thatGis abelian by nilpotent.
2. Results.
First of all recall that ifN denotes the class of all nilpotent groups, then the N-residual of a group G, denoted by GN, is the intersection of all normal subgroupsNofGsuch thatG=N2 N. It is clear thatG=GN 2 N. In the following G0 will denote the derived subgroup of G and G2
<g2:g2G>.
LEMMA. Let GAB be the mutually permutable product of the nil- potent subgroups A and B. Assume that GN is a p-group and that GN A.
Then G satisfies one of the following statements: (a) G2is nilpotent; (b) GN is abelian.
PROOF. Assume the result is not true and G is a minimal counter- example, so thatGis not nilpotent,GN 61 andpis a prime divisor ofG;
also, by supersolubility [1],p>2. We begin with two statements and collect the consequences afterwards.
(i) Op0(G)1.
The conditions of the Lemma carry over to quotient groups. Assume thatOp0(G)M61:ThenM\GN 1 andMbelongs to the hypercenter ofGsinceMandMGN=GN are operator isomorphic. Now the nilpotence of (G=M)2 leads to the nilpotence ofG2M=M and G2; on the other hand, if (G=M)N (GN M)=Mis abelian, so isGN. Thus the statement is true by minimality ofG.
In particular, the Fitting subgroupF(G) ofGis ap-group andpis the biggest prime dividing the order ofG.
(ii) Op(G)G.
LetTbe any normal subgroup ofG; by [6], Lemma 1 (i),(ii) we have that (T\A)(T\B) is a normal subgroup ofGand that it is the product of the two mutually permutable subgroups T\A and T\B. Furthermore jG:(T\A)(T\B)j is a divisor of jG:Tj2. Applying this to the smallest normal subgroup ofp-power indexOp(G)K;we haveK(K\A)(K\B), and forK 6Gwe find that the Lemma is true forK. IfK2is nilpotent, so is G2K2F(G); likewise KN GN and if one is abelian, the other is, too.
Again minimality ofGyields the result.
Now we collect the consequences:F(G) is the Sylowp- subgroupPofG by (i), furthermore F(G)GN by (ii) and therefore F(G)GN G0. Since CG(F(G))F(G) and GN A, we obtain thatA does not possess elements of order prime top, soAGN PandGAHwhereHis the Hallp0- subgroup ofB. MoreoverG=AHis abelian of exponent dividing p 1.
(iii) [A;H]A.
This follows fromA[G;A][H;A]A0[H;A]F(A).
(iv) A\BA0.
By [10, Proposition I(1.7), p. 206] applied to A=A0 and H we obtain A=A0([A;H]A0=A0)CA=A0(H) with trivial intersection of the factors since the p-group is abelian. Now (iii) yields (B\A)A0=A0CA=A0(H)1 and this is (iv).
LetT[A0;A]. We have the following strengthening of (iv):
(v) (A\B)TA0.
First, (A\B)Tis normalized byHand byA. Every subgroupU with (A\B)TUAis normalized by B sinceUUB\A. So Hinduces power automorphisms on A=(A\B)T. By a result of Cooper (see [9, Corollary 2.2.2, p. 339]) we have [A;[A;H]](A\B)T. Now (v) follows from (iii) and (iv).
(vi) IfAis nonabelian, thenG2 is nilpotent.
Choose an elementx61 ofH, it induces a non-identity automorphism on A=A0. Since all subgroups Dwith A0DA are normalized by Band therefore byx, there is a numberssuch thatx 1uxA0usA0for allu2A;
also x 1[u1;u2]xT[us1;us2]T[u1;u2]s2T for all u1;u22A. But x cen- tralizesA0=T, sos21 mod. exp(A0=T) andx 1uxA0u 1A0for allu2A.
Now [x2;A]A0; by a famous result of P. Hall [12, Theorem 7] we have: If the normal subgroupMof the groupTis nilpotent andT=M0is nilpotent,
then T is nilpotent. Applying this to A and hx2;Aiyields that hx2;Aiis nilpotent and in particular thep0- elementx2centralizesA. Sox21 and H21 sincex2Hwas arbitrary. ThereforeG2AH2Ais nilpotent.
We have shown that there is no minimal counterexample. Now the proof is complete.
Notice that under the conditions of the Lemma the following is always true:
(+)GN \B1 if and only ifGN is abelian.
We are now in the position to prove our
THEOREM. Let GAB be the mutually permutable product of A and B with A and B nilpotent. Then(G2)N is abelian. In particular,if G is of odd order we obtain(GN)is abelian.
PROOF. Assume the result is not true and thatGis a minimal coun- terexample. Again the result of Asaad and Shaalan [1] yields thatGis su- persoluble. So we have thatGN G0F(G) andPF(G) ifPis the Sylow p-subgroup ofGwithpthe largest prime divisor of the order ofG. Clearly, every quotient groupG=NofGinherits the hypothesis. We deduce at once
(i) Gpossesses only one minimal normal subgroup.
Assume thatN1 andN2 are two different minimal normal subgroups, then ((G2)N)0Nifori1;2 and ((G2)N)0N1\N2 1.
By (i), we have now that PF(G) and therefore GN G0 PG2G.
It is a celebrated result of Wielandt (see [13, Satz 4.6, p. 676]) that in the product of two nilpotent groupsX;Ythe corresponding Hallp-subgroups permute with each other and that their product is a Hallp-subgroup ofXY, for every set of primesp. In our case we haveP(P\A)(P\B) and the product of the twop-complements ofAandBrespectively gives rise to a p-complementQofG, with (Q\A)(Q\B)Q, andGPQ. The quotient groupG=PQis abelian by (i) and supersolubility ofG.
We study now the subgroup BP. Our intention is to prove that BP satisfies the hypothesis of the Lemma. We consider the intersections ofA andBwithPandQand obtain
(ii) B\QnormalizesA\P.
The subgroup A\P is the Sylow p-subgroup of (B\Q)A
(B\Q)(A\Q)(A\P)Q(A\P), and (ii) follows.
(iii) BP(B\Q)P(A\P)Bsatisfies the conditions of the Lem- ma, taking (A\P) andBinstead ofAandB.
Notice that [B\Q;P] is a normal subgroup of (B\Q)PBP. Now [B\Q;P][B\Q;(A\P)(B\P)][B\Q;A\P]A\Pby (ii). But also (BP)N [B\Q;P]A\P (notice that B\Q acts on P and then P[B\Q;P]CP(B\Q), which shows thatBP=[B\Q;P] is nilpotent). By symmetry we have also
(iv) APsatisfies the conditions of the Lemma, taking there (B\P) andAinstead ofAandB.
(v) APandBPare normal subgroups ofG.
This follows sinceG=Pis abelian.
Now we will exclude all possibilities for the counterexampleG.
(vi) (AP)N can not be abelian.
Assume first that the normal subgroupAPofGis nilpotent. This means APF(G)P andPA(B\P). Let Hbe the Hall p'-subgroup of B.
ThenGN [GN;H][P;H][A;H]A. By our Lemma,GN is abelian orG2is nilpotent, so also (G2)N is abelian.
Now consider the case that (AP)N W is nontrivial and abelian. If (BP)2 is nilpotent, then 1W(AP)(BP)2Gis a sequence of normal subgroups ofGwithWabelian, (AP)(BP)2=Wnilpotent, andG=(AP)(BP)2 of exponent 2. So (G2)N is abelian. If also (BP)N is nontrivial and abelian, we have on one hand by (+) and (iv) W\A1 andW B\P, on the other hand by (+) and (iii), (BP)N \B1 and (BP)N A\P. Now (BP)N \W 1 contrary to the existence of only one minimal normal subgroup.
We come to the conclusion
(vii) There is no counterexample to the Theorem.
By (vi) we know that the only possibility left is that (AP)2and (BP)2are nilpotent. But then (AP)2(BP)2G2is nilpotent. This finishes the proof of the Theorem.
The nilpotent residual of G need not be abelian in the situation con- sidered. In the following we give two examples:
EXAMPLE1. Consider the group B ha; bja3[b; a; a]; b3[b; a; a]2; [b; a]3[b; a; a]3[b; a; b]1i: Bhas order34 and nilpotency class3.
1. All elements of BnB0 have order 9. Notice that (abk)3
a3b3k[b;a;a]k[b;a;a]mwithm12kk1 mod o([b;a;a]). Thus B3 h[b;a;a]i.
2. B0 h[a;b];[b;a;a]i.
3. Let a be the automorphism of B given by aaa2, ba
b2[b;a;a]2b5. Note thatadefines an automorphism of the groupBby realizing that [a;b]a[a2;b5][a 1;b 1] is conjugate to [a;b].
4. Let d[b;a][b 1;a 1]. It is clear by the above remark that da d. Therefore we can considerA hd;ai hdi hai.
LetG[B]haibe the corresponding semidirect product. The groupG is a mutually permutable product ofAandB. Moreover,CoreG(A\B)1 andGN Bis not abelian.
EXAMPLE2. Letpbe an odd prime; denote byPan extraspecial group of orderp2t1 witht1 and of exponentp. It is clear that the subgroup F(P)P0Z(P) has orderp. MoreoverP has nilpotency class2and the quotientP=Z(P) is elementary abelian. Now the application of [10, 20.8 (b), p. 81] yields that there exists an automorphismaof order2ofP such thata fixes the elements ofZ(P) and induces inversion onP=Z(P).
Consider now G[P]haithe corresponding semidirect product. De- noteBPandA ha;Z(P)i. ThenGis the mutually permutable product of A and B both nilpotent. Moreover A\BZ(P) and GN P is not abelian.
Acknowledgments. The authors are indebted to the referee for hints how to improve the transparency of this paper. The first and fifth authors are supported by Proyecto MTM2004-08219-C02-02 (MCyT) and FEDER (European Union).
REFERENCES
[1] M. ASAAD- A. SHAALAN,On the supersolvability of finite groups,Arch. Math., 53(1989), pp. 318±326.
[2] A. BALLESTER-BOLINCHES - J. COSSEY - R. ESTEBAN-ROMERO, On totally permutable products of finite groups,J. of Algebra,293(2005), pp. 269±278.
[3] A. BALLESTER-BOLINCHES- M. D. PEÂREZ-RAMOS- M. C. PEDRAZA-AGUILERA, Totally and mutually permutable products of finite groups, Groups St.
Andrews 1997 in Bath I, pp. 65±68. London Math. Soc. Lecture Note Ser.
260. Cambridge University Press, Cambridge, 1999.
[4] J. C. BEIDLEMAN- H. Heineken,Survey of mutually and totally permutable
products in infinite groups,Topics in infinite groups, pp. 45±62, Quad. Mat.8, Dept. Math. Seconda Univ. Napoli, Caserta, 2001.
[5] J. C. BEIDLEMAN- H. HEINEKEN,Totally permutable torsion subgroups,J.
Group Theory,2(1999), pp. 377±392.
[6] J. C. BEIDLEMAN- H. HEINEKEN,Mutually permutable subgroups and group classes,Arch. Math.,85(2005), pp. 18±30.
[7] A. CAROCCA,p-supersolvability of factorized groups,Hokkaido Math. J.,21 (1992), pp. 395±403.
[8] A. CAROCCA - R. MAIER, Theorems of Kegel-Wielandt type, Groups St.
Andrews 1997 in Bath I, 195-201. London Math. Soc. Lecture Note Ser., 260. Cambridge University Press, Cambridge, 1999.
[9] C. D. H. COOPER,Power automorphisms of a group,Math. Z.,107(1968), pp.
335±356.
[10] K. DOERK- T. HAWKES,Finite Soluble Groups,Walter De Gruyter, Berlin, New York, 1992.
[11] D. GORENSTEIN,Finite Groups,Chelsea Pub. Co., New York, 1980.
[12] P. HALL, Some sufficient conditions for a group to be nilpotent.Illinois J.
Math.,2(1958), pp. 787±801.
[13] B. HUPPERT,Endliche Gruppen I,Springer-Verlag, Berlin, Heidelberg, New York, 1967.
[14] G. ZAPPA,Sui gruppi supersolubili, Rend. Sem. U. Roma IV ser.,2(1938), pp. 323±330.
[15] G. ZAPPA, Remark on a recent paper of O. Ore, Duke Math. J., 6 (1940), pp. 511±512.
Manoscritto pervenuto in redazione il 9 gennaio 2006.